HAL Id: tel-01673833
https://tel.archives-ouvertes.fr/tel-01673833v2
Submitted on 11 Jan 2018
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
In vivo study of the suppression of cell-autonomous and
systemic RNA silencing by the Peanut clump virus
protein P15
Marco Incarbone
To cite this version:
Marco Incarbone. In vivo study of the suppression of cell-autonomous and systemic RNA silencing by the Peanut clump virus protein P15. Virology. Université de Strasbourg, 2016. English. �NNT : 2016STRAJ090�. �tel-01673833v2�
UNIVERSITÉ DE STRASBOURG
ECOLE DOCTORALE DE LA VIE ET DE LA SANTE
THÈSE
Présentée par
Marco INCARBONE
à
L’Institut de biologie moléculaire des plantes, CNRS
pour obtenir le grade de
DOCTEUR DE L’UNIVERSITÉ DE STRASBOURG
Spécialité : Aspects Moléculaires et Cellulaires de la Biologie
In vivo study of the suppression of
cell-autonomous and systemic RNA silencing by the
Peanut clump virus protein P15
Soutenue le 5 Décembre 2016 devant la Commission d’Examen :
Directeur de Thèse
Dr Patrice DUNOYERRapporteur Externe
Dr Hervé VAUCHERETRapporteur Externe
Prof Peter BRODERSENExaminateur Interne
Dr Sébastien PFEFFER“Welcome. And congratulations. I am delighted that you could make it. Getting here wasn’t easy, I know. In fact, I suspect it was a little tougher than you realize.
To begin with, for you to be here now trillions of drifting atoms had somehow to assemble in an intricate and intriguingly obliging manner to create you. It’s an arrangement so specialized and particular that it has never been tried before and will only exist this once. For the next many years (we hope) these tiny particles will uncomplainingly engage in all the billions of deft, cooperative efforts necessary to keep you intact and let you experience the supremely agreeable but generally underappreciated state known as existence. (…)
So thank goodness for atoms. But the fact that you have atoms and that they assemble in such a willing manner is only part of what got you here. To be here now, alive in the twenty-first century and smart enough to know it, you also had to be the beneficiary of an extraordinary string of biological good fortune. Survival on earth is a surprisingly tricky business. Of the billions and billions of species of living things that have existed since the dawn of time, most – 99.99 percent – are no longer around. Life on earth, you see, is not only brief but dismayingly tenuous. It is a curious feature of our existence that we come from a planet that is very good at promoting life but even better at extinguishing it.”
Acknowledgements
First of all, I would like to thank my wonderful parents, Enrico and Karin, for always being patiently there to support, understand and motivate me. Equally, I’d like to thank my brother Roby and all the family in the Incarbone and Judkins/Mee clans. To all the friends from Torino and now in all corners of the globe, thanks guys!
These four years of PhD have been absolutely great, and I owe this mostly to all the beautiful people I have had the pleasure to meet here in Strasbourg. I would like to acknowledge the Garcia family, Adrien Pasquier, Jerome Zervudacki and my PhD brother Thomas Montavon. I’d also like to thank everybody at IBMP, but especially Yerim, Aude, Fabrice, Jean, Nina, Clement, Todd, Kamal, Khalid, Annette, Alyssa, Mattia, Elodie, Salah, Malek, Natka, Ahmed, Vianney, Ritz, David, Eduardo, Lea, Adrien and Thibaut, Clara, Esther, Christophe, the gardeners, Michelle, the administration staff. I would also like to thank CNRS and Labex for financing my works and providing many occasions to discuss with outstanding scientists. I am indebted to Sigrun Reumann, together with Delphine and Piotr, for teaching me the tricky business of peroxisome isolation.
A very special thanks goes to Patrice Dunoyer, who took me under his wing and coached me for four years, always pushing me to do better, to think, to hypothesize, to analyze, to try out new ideas, to try to be a good and thorough scientist. The fact that he continued to do so during some truly brutal times can only go to his credit.
Huge gratitude and affection goes to Marion Clavel for, well, everything.
I’d like to also express my gratitude to Sébastien Pfeffer, Hervé Vaucheret and Peter Brodersen for evaluating this work.
1
TABLE OF CONTENTS
TABLE OF CONTENTS ... 1Figure Index ... 4
INTRODUCTION ... 6
I - Plant viruses and antiviral defense ... 8II - RNA interference ... 12
IIa - Cell-autonomous RNAi: the dicing step ... 13
IIb - Cell-autonomous RNAi: the amplification step ... 16
IIc - Cell-autonomous RNAi: the effector step ... 17
IId - Cell-to-cell and systemic RNAi ... 20
III - Viral suppression of RNA silencing ... 23
IV - Peanut Clump Virus (PCV) and P15, its VSR ... 37
V - Peroxisomes and peroxisomal protein import ... 39
CHAPTER 1 ... 43
EFFECT OF P15 ON CELL- AND NON-CELL AUTONOMOUS RNA SILENCING ... 43
1.1 - Characterization of the mode of action of P15 in the suppression of cell-autonomous RNA silencing ... 451.1a - Forward genetic screens via EMS mutagenesis ... 45
1.1b - AGO immunoprecipitation ... 46
1.1c - P15FHA immunoprecipitation ... 49
1.2 - Characterization of the mode of action of P15 in the suppression of non-cell autonomous RNA silencing ... 52
1.3 - Antiviral siRNA population generated during PCV infection ... 54
1.4 - Conclusion ... 56
CHAPTER 2 ... 61
ROLE OF P15 PEROXISOMAL LOCALIZATION IN PCV SYSTEMIC MOVEMENT ... 61
2.1 - Isolation of peroxisomes from plants expressing p35S:P15 ... 632.1a – Non-infected and TRV-PDS-infected plants ... 64
2.1b - TuMV-GFP-infected and TuMV-GFP AS9-infected plants ... 65
2.2 - Isolation of peroxisomes from PCV-infected plants ... 67
2.3 - Significance of siRNA peroxisomal import in PCV systemic movement ... 69
2 2.4 - Significance of phloem loading/unloading of siRNA in the systemic restriction of PCVΔN6 ... 70
2.5 - Conclusions ... 72
CHAPTER 3 ... 76
PEROXISOMAL TARGETING AS A TOOL TO IDENTIFY NOVEL MOLECULAR
INTERACTIONS ... 76
3.1 – Experimental system design ... 783.2 - Peroxisomal targeting of endogenous RNAi factors: DRB4 and AGO2 ... 79
3.3 - Peroxisomal targeting of viral suppressors of RNA silencing: P38 and P19 ... 81
3.4 - Conclusions ... 83
DISCUSSION AND PERSPECTIVES ... 86
4.1 – P15, a compelling example of viral evolution ... 874.2 – Peroxisomal confinement of host defense factors could be used by other pathogens ... 89
4.3 – Tools for future research ... 90
4.3a – DCL2 biology and DCL2/DCL4 interplay during antiviral defense ... 90
4.3b – Uncoupling of systemic RNAi from cell-autonomous RNAi during unbiased viral infection ... 91
4.3c – Peroxisomal isolation to uncover labile interactions and to investigate viral replication complexes ... 92
MATERIALS & METHODS ... 94
5.1 - Cloning and transgene construction ... 955.2 - A. thaliana stable transformation and transgenic line selection ... 96
5.3 - Plant germination and growth ... 96
5.4 - Virus infection and patch assays ... 97
5.5 - EMS mutagenesis ... 99
5.6 - Immunoprecipitation ... 99
5.7 - Peroxisome isolation ... 100
5.8 - PCV purification ... 102
5.9 - RNA extraction ... 104
5.10 - Protein extraction ... 105
5.11 - Northern blotting ... 105
5.12 - Western blotting ... 107
5.13 - Immunolocalisation ... 107
3 5.14 - Mass spectrometry proteomic analysis ... 108
5.15 – Primers ... 109
ACRONYMS & ABBREVIATIONS ... 114
REFERENCES ... 117
ANNEXES ... 133
Supplementary Figures: Annex 1Résumé de thèse en français: Annex 2 Communication at a congress: Annex 3 Manuscript submitted to a journal: Annex 4
4
Figure Index
Figure I: Plant virus families and genera (page 8)
Figure II: Cell-autonomous antiviral RNAi in plants (page 13) Figure III: Systemic antiviral RNAi in plants (page 22)
Figure IV: The peroxisomal importomer (page 40)
Figure 1.1: The SUC:SUL RNAi reporter system (SS) (page 44)
Figure 1.2: Mutagenesis of p35S:P15/SS to obtain SS phenotype revertants (page 45) Figure 1.3: Effect of P15FHA on sRNA loading into AGO1 and AGO2 (page 46)
Figure 1.4: Effect of exclusive DCL2 siRNA generation on P15FHA VSR activity (page 48) Figure 1.5: Association of P15FHA with sRNA (page 49)
Figure 1.6: Effect of exclusive DCL2 activity versus DCL2 absence on P15FHA association
with sRNA (page 50)
Figure 1.7: Effect of P15FHA on cell-to-cell movement of 21nt and 22nt siRNA (page 52) Figure 1.8: Antiviral siRNA population during PCV infection (page 54)
Figure 1.9: P15FHA sRNA sequestration hierarchy in vivo (page 56)
Figure 2.1: Effect of peroxisomal localization of P15 on PCV infection and on
cell-autonomous RNAi (page 62)
Figure 2.2: Analysis of peroxisomes isolated from P15-expressing plants (page 64)
Figure 2.3: Effect of siRNA sequestration by another VSR on P15wt-dependent
piggybacking of siRNA into peroxisomes (page 65)
Figure 2.4: Analysis of peroxisomes isolated from PCV-infected plants (page 67)
Figure 2.5: Relevance of P15wt-dependent siRNA peroxisomal import in PCV movement
(page 69)
Figure 2.6: Relevance of 21nt versus 22nt siRNA peroxisomal import in PCV movement
5
Figure 2.7: Relevance of antiviral siRNA phloem loading/unloading in PCVΔN6 systemic
restriction (page 71)
Figure 3.1: PTS1 addition to DRB4 and AGO2 to trigger peroxisomal targeting (page 79) Figure 3.2: Effect of AGO2 overexpression on endogenous sRNA homeostasis (page 80) Figure 3.3: PTS1 addition to P38 and P19 to trigger peroxisomal targeting (page 82)
Figure 3.4: Advantages of P15 and P19 peroxisomal targeting in the identification of
molecular interactions, compared to immunoprecipitation (page 85)
Figure 4.1: PCVΔN6 movement is blocked by systemic antiviral RNAi, which is triggered
by phloematically mobile 21nt siRNA (page 89)
Figure 4.2: PCVwt prevents movement of siRNA by confining them in peroxisomes
6
7
The year 1892 was an important one in biology. In an age when people were just starting to accept the fact that diseases were caused by specific and transmissible microscopical living things and not by “miasmas”, the Russian botanist Dmitri Ivanovsky made a truly sensational discovery, that would profoundly change biology and cause more than one headache to whoever has tried since then to provide a univocal and unambiguous definition of life. Ivanovsky observed that the agent causing tobacco mosaic disease was able to pass undisturbed through filters that retained bacteria, indicating that it was smaller than the smallest known living creature. Although at the time he didn’t have the tools to describe the true nature of what he had found, and as often happens in science he then proceeded to follow numerous valid yet ultimately false leads, Ivanovsky had discovered viruses.
Since then, more than a century of ingenuity, patient experimentation and stubborn investigation have yielded priceless evidence on what viruses are and how they go about the business of surviving and adapting to an ever changing world while not possessing the essential machinery to do so. Outside their host they are inorganic matter, not very different from a strand of hair or a microscopic grain of sand. However, when they come into contact with their host in the right conditions, viruses spring to life in ways that are fascinating to observe even through the indirect and partial view afforded by modern molecular biology. They hijack proteins, nucleic acids and membranes, they replicate, they mutate, they migrate, they fight efficient and diversified defensive host reactions, they board specific vectors and change their behavior to increase their chances of successfully reaching their next host. They achieve all this, and much else, mostly by using the host’s machinery for their own purposes, by warping the host’s vital mechanisms to fulfill their own agenda. Viruses are, in a way, the ultimate manipulators.
The main goal of this experimental work is to provide a clear snapshot of one such manipulation and place it in its context.
Figure I: Plant virus families and genera.
Plant virus families (suffix –viridae) and genera (suffix –virus). They are here divided according to the nature of their genome. The shape of the virion is depicted for each genus. Source: http://www.emporia.edu
8
I - Plant viruses and antiviral defense
A virus is defined as “a set of one or more nucleic acid template molecules, normally encased in a protective coat or coats of protein or lipoprotein, that is able to organize its own replication only within suitable host cells. It can usually be horizontally transmitted between hosts. Within such cells, virus replication is (1) dependent on the host’s protein-synthesizing machinery, (2) organized from pools of the required materials rather than by binary fission, (3) located at sites that are not separated from the host cell contents by a lipoprotein bilayer membrane, and (4) continually giving rise to variants through various kinds of change in the viral nucleic acid” (Hull, 2002).
The general absence of proof-reading ability in viral replicases, leading to a high rate of mutation, coupled to the vast number of replication events taking place during a typical infection, have in time generated an almost boundless variety of shapes, strategies and tricks that viruses employ to successfully complete their infectious cycle. Broadly speaking, viruses are classified according to the nature of their genomic nucleic acid (DNA or RNA, single-strand or double-strand, coding or complementary strand) and the shape of their capsid. Other parameters of classification include replication strategy, host range and transmission vectors, among others. They are taxonomically classified into families, genera, species and strains, following a nomenclature that is different from the classical Linnaean binomial one used for all other forms of life. This nomenclature uses English and allows univocal abbreviation of virus names into acronyms to facilitate use (e.g. Tobacco Mosaic Virus becomes TMV, Turnip Mosaic Virus becomes TuMV, and so on). We shall here focus on plant viruses (Fig. I) and their interaction with and suppression of host antiviral mechanisms. Plant DNA viruses count among their ranks species that are of great agricultural importance given the yield loss caused yearly, such as those belonging to the genus Geminivirus. However, the vast majority of plant viruses have an RNA genome. Of these, most species possess a single-stranded so-called (+)-polarity genome ((+)-ssRNA), meaning that the genomic RNA is the coding strand and can be directly translated into protein. The protein that is the object of this work is encoded by one such virus, Peanut Clump Virus. While our knowledge on plant RNA virus life cycles and strategies is the object of many detailed texts
9
and reviews, for the sake of fluidity it will here suffice to provide a very broad and largely simplistic overview of the typical life cycle of a plant (+)-ssRNA virus. Given the topic, there are probably several exceptions for each of the following statements.
A virus first enters its host as a virion (RNA + capsid) through direct contact with an infected host or through a vector. Upon entry the capsid (made up of coat protein) is removed and the genomic RNA is translated by ribosomes into key factors such as the replicase. The replicase is a viral RNA-dependent RNA-polymerase (RDRP), that uses the genomic RNA as a template to synthesize the (-) complementary strand. This step generates a double-stranded RNA that is probably short-lived, yet of paramount importance for the host (and of great danger for the virus) as it is a potent elicitor of RNA interference, a staple mechanism in plant antiviral defense. The freshly made (-) strand can then be used to synthesize more (+) RNA, that can in turn be translated into protein, copied, or encapsidated in newly generated coat protein to form new virions. Replication of (+)-ssRNA viruses usually takes place in the cytoplasm within membranous invaginations induced by the virus on specific subcellular compartments or organelles (reviewed in Grangeon et al., 2012). Once the cells of entry are colonized, viruses proceed to move to neighboring cells through plasmodesmata (reviewed in Heinlein, 2015) and ultimately to distant tissues in the plant through the phloem (reviewed in Hipper et al., 2013), in the form of virions or dedicated ribonucleoproteic (RNP) complexes, depending on the species, thereby colonizing new cells. As the last step of an infectious cycle the virus must reach a new host, most often by being acquired by a specific vector (ranging from fungi to insects)(reviewed in Blanc et al., 2011). While typically encoding 5 to 10 essential and often multifunctional proteins, viruses largely rely on host proteins to complete their life cycle (reviewed in Hyodo et al., 2014; Wang, 2014).
There are several reported cases of symbioses between viruses and their hosts, and several instances are described in which viruses provide their plant hosts with drought tolerance, cold tolerance and other advantages, though how they achieve this remains unclear (reviewed in Roossinck, 2011). Modern meta-genomic approaches have revealed many persistent viruses that cause no visible disease and may have co-evolved with their hosts for millennia. Despite this, and possibly because scientific enquiry has mostly focused on the
10
most evident, pathogenic and economically detrimental plant-virus interactions, viruses are generally considered selfish genetic elements that thrive at the expense of their hosts. Plants have thus evolved several layers of defense to hinder the establishment and propagation of viral infections.
The so-called R-gene (incompatible, or gene-for-gene) response, which is mostly virus-specific and results from prolonged virus-host coevolution, consists in the recognition of a viral avirulence (Avr) factor by a single dominant resistance (R) gene product (reviewed in de Ronde et al., 2014; Mandadi and Scholthof, 2013; Dodds and Rathjen, 2010). This kind of response can be triggered by viruses, bacteria or fungi. The majority of these R genes belong to the NB-LRR family. Triggering of R genes generally leads to a programmed cell death response that is quite rapid (3 or 4 days post-infection) and contains the spread of the virus by killing the infected tissue. This hypersensitive response (HR) entails dramatic metabolic changes in the synthesis of hormones such as salicylic acid (SA), jasmonic acid (JA), ethylene and nitric oxide, the accumulation of reactive oxygen species, calcium ion influx, callose deposition at plasmodesmata, modification of membrane permeability and expression of pathogenesis-related (PR) proteins (Mandadi and Scholthof, 2013; Pallas and García, 2011). Avr/R protein interactions can also initiate systemic acquired resistance (SAR), whose exact mode of action remains unclear though it has been associated to an accumulation of SA and JA in tissues distant from the site of infection (Mandadi and Scholthof, 2013). SAR doesn’t cause cell death, and provides a long-lasting systemic immune response against subsequent infections.
While dominant resistance has received a great deal of attention, there are other less investigated forms of antiviral defense. For example, A. thaliana proteins RTM1, RTM2 and RTM3 have been shown to restrict the movement of Tobacco Etch Virus (TEV) and other Potyviruses independently of HR/SAR through recognition of the 5’ end of the coat protein (Chisholm et al., 2001; Decroocq et al., 2009). JAX1, a lectin protein closely related to RTM1, has been shown to inhibit Plantago Asiatica Mosaic Virus (PLAMV) replication (Yamaji
et al., 2012). Lectin proteins could therefore represent another barrier against viruses,
11
Yamaji et al., 2012). Plants can also employ the ubiquitin proteasome pathway to target viral proteins for degradation, as for example happens to the movement proteins of Tobacco Mosaic
Virus (TMV) and Turnip Yellow Mosaic virus (TYMV), thereby decreasing virulence and
pathogenicity (Drugeon and Jupin, 2002; Reichel and Beachy, 2000). It has been recently shown that upon recognition of begomovirus NSP protein, A. thaliana NIK1 (a LRR receptor-like kinase) triggers global translational suppression (Zorzatto et al., 2015). The authors of this study found that NIK1 activation leads to LIMYB-mediated downregulation of ribosomal protein gene transcription. This in turn results in inhibition of ribosomal protein synthesis, decreased association of viral mRNA with polyribosomes and increased tolerance to the virus. Plant RNA quality control pathways have recently been proven to contribute to antiviral defense. UPF1, a key player in non-sense mediated decay (NMD), restricts Potato Virus X (PVX-GFP) by recognizing the internal termination codons and long 3’ UTRs present in its subgenomic RNAs (Garcia et al., 2014). UPF1 also restrains the accumulation of Turnip Crinkle Virus (TCV), and could constitute a limiting factor to all the RNA viruses that, because of their genomic organization and expression strategy, sport sequences recognized as aberrant by the NMD machinery. Finally, A. thaliana RTL1, an RNAse-III enzyme that processes dsRNA, has been shown to be strongly upregulated during viral infection and is likely to play a role in antiviral defense (Shamandi et al., 2015). While the impossibility of obtaining a knock-out mutant of RTL1 made it difficult for the authors to assess the exact implication of this factor in antiviral defense, they did show that its overexpression causes increased virus accumulation during infections by TVCV, but not TCV, CMV and TYMV. Though in the case of TVCV this result may seem counterintuitive, RTL1 was shown to act upstream of RNA interference, the main antiviral defense in plants. Therefore, RTL1 overexpression likely entails depletion of substrate for RNA interference and subsequent impairement of this pathway.
Since experimentally dissecting the molecular events characterizing each infection often represents a truly daunting challenge, it is very likely many offensive and defensive pathways remain to be uncovered.
With the exception of NMD and the case of RTL1, the defensive mechanisms listed above involve species-specific reactions, effectors or patterns, so it can be speculated that mutation
12
by a given virus may allow it to evade these defenses. It is a testimony to the brilliance of plant evolution that the main antiviral mechanism targets the universal and unavoidable product of RNA virus replication: double-stranded RNA. This conserved and multi-layered mechanism, that attacks double-stranded RNA engendered by viral genomic RNA replication or secondary structures and generates sequence-specific defense elicitors, is called RNA interference (RNAi), but is also known as RNA silencing or post-transcriptional gene silencing (PTGS).
II - RNA interference
Most eukaryotes possess RNA interference pathways that are remarkably similar across kingdoms. Although prokaryotes do not encode proteins analogous to eukaryotic silencing factors, they employ mechanisms that are functionally analogous. This structural and functional conservation coupled to frequent functional redundancy strongly suggests that tight control of RNA has been a priority since very early in the history of life. If we briefly allow ourselves to speculate on events in very ancient times, it is easy to infer how RNAi-like mechanisms may have represented a milestone in the evolution of complex life. According to the widely endorsed “RNA world” theory, at the dawn of life on Earth RNA used to be a dominant organic molecule, able to self-replicate and acting not only as vehicle of information but also as reaction catalyzer (ribozyme). These independent or semi-independent RNA molecules, that were presumably more or less structured, and possibly formed distinct populations of nearly identical individuals, may have been the main form of life for a very long time. In all likelihood these RNAs may have represented a significant obstruction for the rise of any new, more complex form of life possibly implementing proteins, if not DNA, in its molecular functioning. The development of machinery able to target replicating or structured RNA in a sequence-specific manner would have conferred a tremendous evolutionary advantage to its bearer. In this scenario, the emergence of proto-RNAi (and possibly other forms of RNA control) may have marked the beginning of the end for the “RNA world”, causing the gradual yet inexorable extinction of most forms of independent RNA and issuing the beginning of a new world order in which RNA came to be merely one player in the intricate workings of a cell. This formidable evolutionary pressure
Figure II: Cell-autonomous antiviral RNAi in plants.
Schematic representation of cell-autonomous antiviral RNAi in plants. Upon entry into a cell, a ssRNA virus replicates, generating double-stranded RNA (1-2). This is processed by DCL enzymes into 21-22nt siRNAs (3-4), which are then incorporated into AGO proteins (5-6) to mediate the sequence-specific antiviral response (7). At the top is a schematic representation of the main endogenous RNAi pathways, such as genome maintenance and transcriptional regulation through 24nt siRNAs, or the regulation of mRNA accumulation and translation through 21-22nt miRNAs and ta-siRNAs.
13
would have only allowed the survival of those independent RNAs that managed to protect themselves with a shell of some kind and evade or counteract RNAi, learning at the same time how to take advantage of the increasingly complex machinery employed by the forms of life they inhabited. Testimony to this possible turn of events is the fact that viroids, the only known form of independent RNA that is non-coding and not encapsidated (excluding some active non-coding transposable elements), are present only in plants as a handful of species, perhaps an elusive reminder of what life was like in the very distant past.
RNAi was first observed by investigators working on plants, but it was Andrew Fire, Craig Mello and colleagues working on C. elegans that provided evidence that it is a reaction to the presence of double-stranded RNA and entails potent gene silencing (Fire et al., 1998). In plants RNAi plays a key role in many aspects of life, from gene regulation to genome maintenance to the control of invading or aberrant RNA. In a nutshell, the long dsRNA trigger is processed by Dicer-like enzymes (DCLs) through endonucleolytic activity into 21-24nt small RNA (sRNA), one strand of which is loaded into an Argonaute (AGO) effector that uses it as a template to recognize complementary single-stranded RNAs and cleave them or inhibit their translation. In specific conditions, dsRNA can also be synthesized from ssRNA templates by RNA-dependent RNA-polymerases (RDRs), leading to further generation of sRNAs by DCLs and amplification of the silencing reaction in what is generally known as transitivity. This overview will outline the basic workings of RNAi in plants, focusing on its role in defense against RNA viruses (Fig. II).
IIa - Cell-autonomous RNAi: the dicing step
In the model plant specie A. thaliana, the four Dicer-like (DCL) enzymes and their products have been well characterized (reviewed in Bologna and Voinnet, 2014, Fukudome and Fukuhara, 2016), although it remains mostly unclear how they recognize/discriminate their substrates and if they are aided/chaperoned or inhibited by other proteins. DCL RNAse III enzymes possess several domains: a DExD-box, a helicase-C, a domain of unknown function, a PIWI/ARGONAUTE/ZWILLE (PAZ), two RNAse III domains and two dsRBDs (except DCL2 that has one)(Margis et al., 2006). They process long dsRNA into sRNA duplexes with distinctive 2-nt overhangs and 5’ monophosphates in an ATP-dependent
14
fashion (Tang et al., 2003). The length of the α-helix separating the PAZ from the RNAse domains determines the length of the sRNA produced (MacRae et al., 2007). Once generated, all sRNAs are 2’OH-methylated on their 3’ end by HEN1 and thereby protected from degradation (Yu et al., 2005). DCL proteins can generate two main types of sRNA: microRNA (miRNAs, mostly DCL1-dependent) and small interfering RNA (siRNAs, DCL2/DCL3/DCL4-dependent)(Henderson et al., 2006; Xie et al., 2004).
DCL1 processes genomic non-coding imperfectly paired foldback precursors into 21-22nt miRNAs (Park et al., 2002; Reinhart et al., 2002), and is mostly localized in the nucleus (Song et al., 2007). DCL1-dependent miRNAs regulate many genes involved in housekeeping and development, which explains why knocking out DCL1 results in embryonic lethality. While the majority of miRNAs are 21nt long and regulate endogenous protein production mostly through AGO1 by cleaving (or “slicing”) the corresponding mRNAs (German et al., 2008) or inhibiting its translation (Brodersen et al., 2008), a few miRNAs (22nt-long miR173 and miR828, 21nt-long miR390) are capable of triggering RDR6-dependent synthesis of dsRNA from non-coding TAS transcripts, which are then processed by DCL4 into precisely phased trans-acting siRNAs (ta-siRNAs) that proceed to mediate further gene silencing (Allen et al., 2005; Vazquez et al., 2004; Yoshikawa et al., 2005).
DCL3 processes nearly perfectly complementary POLIV/POLV/RDR2-dependent dsRNA originating from transposons and repeats into 24nt siRNAs that mediate RNA-directed DNA methylation (RdDM) through AGO4 (Law et al., 2010; Pontes et al., 2006), that in turn leads to transcriptional gene silencing (TGS). This process is mostly nuclear and nucleolar. 24nt siRNAs generated by DCL3 are also capable through the same pathway of mediating antiviral defense against DNA viruses such as Geminiviruses (Raja et al., 2010). DCL3 is able to process long dsRNA derived from exogenous hairpin constructs, but the 24nt products are not able to mediate effective PTGS (Dunoyer et al., 2007). While DCL1 and DCL3 are capable of generating sRNAs from RNA viruses, to our knowledge their contribution to the defense against these viruses is negligible (Deleris et al., 2006; Xie et al., 2004).
15
While playing important roles in endogenous RNAi pathways and being able to take over in the establishment of TGS upon knock-out of DCL3 (Henderson et al., 2006), DCL4 and DCL2 are the main dicers involved in defense against RNA viruses. They have been shown to act redundantly both in antiviral and endogenous RNAi, DCL2 taking over upon knockout of DCL4 (Deleris et al., 2006). For this reason DCL2 is mostly considered to be a surrogate of DCL4. However, recently published results (Parent et al., 2015) and results presented in this work suggest that DCL2, while undoubtedly serving as an alternate to DCL4, may have more complex roles in antiviral RNAi that have yet to be revealed.
DCL4 processes long perfectly or near-perfectly complementary dsRNA into 21nt siRNAs. It’s responsible for the generation of the aforementioned RDR6-dependent ta-siRNAs, a few “young” miRNAs (e.g. miR822)(Rajagopalan et al., 2006), the processing of endogenous inverted-repeats (IRs) (Zhang et al., 2007), of transgenically delivered inverted-repeats (Dunoyer et al., 2005) and, of greater interest here, the processing of viral RNA. DCL4 has been shown by many different independent studies to be the main DCL in RNAi against RNA viruses and the primary producer of antiviral siRNAs (Bouché et al., 2006; Deleris et
al., 2006a; Garcia-Ruiz et al., 2010; X.-B. Wang et al., 2011). It has been shown in vitro to
preferentially cleave long dsRNA, and to cleave blunt, 1nt- or 2nt-overhangs with similar efficiency (Nagano et al., 2013). DCL4 has one known cofactor: DRB4. DRB4 is required for DCL4 activity in vitro (Fukudome et al., 2011), while in vivo it strongly enhances, but is not mandatory for, DCL4-dependent processing of exogenous inverted-repeats and TAS dsRNA (Dunoyer et al., 2007). While it assists DCL4 in defense against RNA viruses, it also plays antiviral roles through other pathways (Jakubiec et al., 2012; Qu et al., 2008; Zhu et al., 2013).
DCL2 processes long perfectly complementary dsRNA into 22nt siRNAs. Compared to the other DCLs it has been far less extensively studied. It can act redundantly to DCL4 and downstream of RDR6 in the ta-siRNA pathways and in transgene-driven transitivity (Dunoyer et al., 2007; Moissiard et al., 2007). In phloem companion cells the processing of exogenous inverted repeats in the absence of DCL4 is carried out by DCL3, while DCL2 takes over only if both DCL3 and DCL4 are absent (Dunoyer et al., 2007). When acting as surrogate to DCL4 in the processing of ta-siRNAs and exogenous IRs, DCL2 generates
16
significantly less siRNAs than DCL4 would, although this may simply be due to lower DCL2 accumulation. On the other hand, some endogenous IRs (e.g. IR71) are predominantly processed by DCL2 for unknown reasons. In the absence of DCL3 and DCL4, DCL2 acts in an antagonistic fashion toward DCL1 in the production of sRNAs, and has deleterious effects on development in the absence of DCL1 and DCL4 (Bouché et al., 2006). As mentioned above, in the context of an RNA virus infection DCL2 has until now been reported to be activated mostly in dcl4 knock-out conditions (Bouché et al., 2006; Deleris et
al., 2006; Garcia-Ruiz et al., 2010) or, as in the case of TCV infection, where accumulation of
DCL4 and DCL3 are strongly reduced by the virus (Azevedo et al., 2010; Qu et al., 2008). In these cases, and in contrast with ta-siRNA and exogenous IR processing, DCL2 generates siRNAs in amounts comparable to DCL4. However, given that viral titer is increased, these 22nt siRNAs likely do not mediate RNAi as efficiently as 21nt siRNAs. No DCL2 cofactors are known.
IIb - Cell-autonomous RNAi: the amplification step
During a viral infection DCL4 and DCL2 can process not only double-stranded products of the viral RDRP but also products of host-encoded RDR proteins to produce more siRNAs, dubbed secondary siRNAs. This process, known as transitivity, is readily employed by plants on transgenes (Moissiard et al., 2007; Voinnet, 2005) but tightly controlled when it comes to endogenous genes. Only three out of six A. thaliana RDR proteins have an experimentally proven biological role: RDR1, RDR2 and RDR6. RDR1 and RDR6 have been implicated in antiviral RNAi directed against RNA viruses (Wang et al., 2010), and RDR6 has been shown to often play a pivotal role in it (Qu et al., 2008; Wang et al., 2011). Although the role of RDR6 in the amplification and spread of transgene-directed RNAi in reporter systems is well characterized (Moissiard et al., 2007; Voinnet, 2005), the precise trigger, substrate and conditions of antiviral RDR1/RDR6 intervention are not clear. It is not clear, for example, whether RDR generation of viral dsRNA must be primed by a primary siRNA (as miRNAs prime endogenous ta-siRNA generation) or not. Although 21nt sRNAs can trigger RDR activity and transitivity (Moissiard et al., 2007), recent studies suggest that both in endogenous and transgene-triggered RNAi the initiator of RDR amplification is mostly 22nt sRNAs (Chen et al., 2010; Manavella et al., 2012; Parent et al.,
17
2015). Consequently, DCL2-dependent 22nt siRNAs could initiate a putative RDR-dependent amplification step of antiviral reactions. On the other hand, one study has reported that siRNAs produced by DCL2 downstream of RDR6 were not proficient in mediating efficient antiviral RNAi (Wang et al., 2011).
The evidence available today suggests that in A. thaliana there are some general trends in the generation of antiviral primary and secondary siRNAs, but also that there are likely many differences in the mode and timing of attack by DCLs and RDRs between one virus and the other. These differences could be due to intrinsic features of the silencing factors themselves, to the interaction of the virus with the cell machinery, or a combination of the two.
IIc - Cell-autonomous RNAi: the effector step
Although processing by DCL4 and DCL2 certainly reduces viral accumulation to a certain extent and possibly recruits other defensive factors onto the viral replication complexes, this first step of RNAi taken alone is not likely to have meaningful impact on viral accumulation. The fact that A. thaliana encodes ten often redundantly functioning AGO proteins that act downstream of DCLs makes it almost impossible to genetically isolate the contribution of DCL enzymes alone to the antiviral reaction. AGO proteins carry out the so-called effector step of RNAi, the sequence-specific recognition and attack of single-stranded RNAs that are complementary to the sRNAs generated by DCL enzymes. Our knowledge on plant AGO proteins and their links to other eukaryotic Argonautes is extensively reviewed in Poulsen et
al., 2013.
The key ability of AGO proteins is to bind sRNAs, since it allows them to recognize complementary target RNA in a sequence-specific manner. Of the sRNA duplex generated by DCLs, one strand is loaded into AGO (the guide strand), while the other is degraded (the passenger strand, or * strand). In one reported exception in plants, while the guide strand is loaded into AGO1, the passenger strand is loaded into AGO2 (Zhang et al., 2011). In
Drosophila, this last mechanism seems to be widespread (Okamura et al., 2009). Little is
known in plants about the precise cofactors (if any) that are involved in AGO loading and strand discrimination, or about how sRNAs pass from DCLs to AGOs.
18
Many AGOs possess RNAse H-like endonucleolytic activity that allows them to “slice”, thereby destroying, RNA complementary to the sRNA they are carrying (Llave et al., 2002). This RNAseH-like activity also allows AGO1 to remove the passenger strand of siRNA duplexes (Iki et al., 2010) and to prime RDR activity and secondary siRNA generation (Carbonell et al., 2012). All AGO proteins contain four common domains: an N-terminal domain, a PIWI/ARGONAUTE/ZWILLE (PAZ) domain, a middle MID domain and a PIWI domain. The sRNA is bound to the MID (5’) and PAZ (3’) domains (Ma et al., 2004; Song et
al., 2003), while the PIWI domain carries the RNAse catalytic site (Song et al., 2004). While
in animals many proteins have been shown to interact with Ago to form the RNA-induced silencing complex (RISC), such a complex still evades detection in plants despite the considerable efforts made to identify it. In the wake of studies in animals and yeasts, though, some factors have been shown to interact with AGO proteins through the conserved GW motif (Azevedo et al., 2010; El-Shami et al., 2007; Till et al., 2007). Other proteins that have been found to interact with AGO1 and mediate its functioning are TRN1, HSP90 and CYP40 (Cui et al., 2016; Iki et al., 2012; Iki et al., 2010).
Different AGO proteins are loaded with different categories of sRNA, depending on the pathway involved and on the 5’ base of the sRNA. AGO4, AGO3, AGO6 and AGO9 are loaded with DCL3-dependent 24nt siRNAs to mediate TGS (Havecker et al., 2010; Zhang et
al., 2016). AGO7 is involved in TAS3 processing (Montgomery et al., 2008; Adenot et al.,
2006), while AGO10 is involved in specific meristematic miRNA-driven gene regulation and in AGO1 homeostasis (Mallory et al., 2009; Zhu et al., 2011). AGO7 and AGO10 have also been shown to play minor roles in antiviral RNAi (Garcia-Ruiz et al., 2015; Qu et al., 2008). While AGO5 has been associated to antiviral defense against Potato Virus X (PVX), Turnip
Mosaic Virus (TuMV) and Cucumber Mosaic Virus (CMV)(Brosseau and Moffett, 2015;
Garcia-Ruiz et al., 2015; Takeda et al., 2008) the main AGOs responsible for antiviral defense are AGO1 and AGO2 (Carbonell and Carrington, 2015; Pumplin and Voinnet, 2013). In a similar fashion to DCL4 and DCL2, AGO1 and AGO2 can act hierarchically and redundantly in antiviral RNAi, with AGO2 taking over defense in infections that suppress AGO1 function (Harvey et al., 2011). However, several studies suggest that they can operate
19
independently and non-hierarchically, each one acting against a particular virus or in a particular tissue (Garcia-Ruiz et al., 2015; Ma et al., 2015; Wang et al., 2011).
AGO1 is the most important and multifunctional AGO protein, so central to cellular gene regulation that its loss is lethal. AGO1 is by far the main effector in miRNA-mediated gene regulation, operating through both target slicing and translational inhibition. AGO1 also triggers the generation of phased ta-siRNAs through RDR6/DCL4 after being loaded with specific miRNAs (Vazquez et al., 2004; Allen et al., 2005; Yoshikawa et al., 2005; Montgomery et al., 2008). Through the use of an AGO-interacting suppressor of silencing, AGO1 has been shown to be present in two distinct pools, one loaded with miRNAs, the other with siRNAs (Schott et al., 2012). AGO1 is a potent antiviral effector, and while it has been experimentally shown to be involved in defense against Brome Mosaic Virus (BMV), CMV, Turnip Crinkle Virus (TCV) and TuMV (Carbonell and Carrington, 2015; Garcia-Ruiz et
al., 2015; Morel et al., 2002; Qu et al., 2008), it can be assumed that it plays a role in most
RNA virus infections. AGO1 was shown to be responsible for virus-induced gene silencing (VIGS) of endogenous genes in infections where AGO2 was attacking the virus (Ma et al., 2015). One thing that is not clear is whether AGO1 slices viral RNAs or inhibits their translation, or both. AGO1 is predominantly loaded with sRNAs possessing a 5’ terminal uridine (Mi et al., 2008).
While mediating silencing by a few select miRNAs (Maunoury and Vaucheret, 2011; Zhang
et al., 2011) and being recruited to repair DNA double-strand break (Wei et al., 2012), the
preponderant function of AGO2 is thought to be antiviral. Evidence obtained in several labs confirms AGO2 involvement in infections by CMV, PVX, Tobacco Rattle Virus (TRV), TCV and TuMV on A. thaliana (Carbonell et al., 2012; Garcia-Ruiz et al., 2015; Harvey et al., 2011; Jaubert et al., 2011; Ma et al., 2015; Wang et al., 2011), and by Tomato Bushy Stunt Virus (TBSV) on N. benthamiana (Scholthof et al., 2011). The slicing activity of AGO2 is crucial for defense against TuMV (Carbonell et al., 2012), although it is not known if in infection by other viruses it can also act as a translational inhibitor. As for AGO1, given the number of infections in which AGO2 plays an active role, it can be reasonably assumed that it mediates RNAi against all or most RNA viruses. In addition to its direct role in the cleavage of viral
20
RNA, AGO2 mediates a boost in antiviral defense by mediating widespread silencing of endogenous genes, at least in the case of CMV and TuMV infection (Cao et al., 2014), by means of the host-encoded, DCL4/RDR1-dependent virus-activated siRNAs (vasiRNAs). AGO2 is preferentially loaded with 5’ adenosine sRNAs (Mi et al., 2008).
From this summary of the cell-autonomous aspects of antiviral RNAi it is clear that most of the main players are (more or less) known. However, we are sorely missing a more detailed picture, and not because of lack of will but because of the technical challenge involved in observing the molecular details of a DCL/AGO assault on a viral RNA. The tools used in the last 20 years to unravel the functioning of RNAi are elegant inventions that allowed groundbreaking discoveries, but are not sufficient to answer some more specific questions. Investigators need to come up with new and subtler experimental tools and reporter systems to shed light on this next layer of complexity. Concerning antiviral RNAi many unanswered (and difficult to answer) questions come to mind. For example, (i) whether AGOs act as inhibitors of translation (as in the case of miRNA-targeted mRNA), and possibly of replication and encapsidation, (ii) whether DCLs and AGOs attack viruses together, with the aid of specific cofactors, and if there is any cross-talk between them to better coordinate defenses for the specific virus that is invading the plant, or (iii) different stages of the virus life cycle could correspond to different RNAi reactions. Additionally, it would be interesting to further characterize if and how RNAi is linked to other RNA quality control pathways, and what role RNAi plays in viral host range determination. Trying to answer these and many other questions will constitute a great challenge in the future.
IId - Cell-to-cell and systemic RNAi
Even before the precise factors involved in RNAi had been identified it was known that an RNA silencing trigger could move through the phloem from a silenced rootstock to a non-silenced scion, causing the latter to undergo PTGS (Palauqui et al., 1997; Voinnet et al., 1998). Moreover, it was clear that this silencing signal was capable of moving cell-to-cell, presumably through plasmodesmata (Himber et al., 2003; Voinnet and Baulcombe, 1997). Taken together these works suggested that the sequence-specific silencing signal, once induced, moves out of the incipient cells through plasmodesmata, reaches the phloem and
21
enters it, and upon reaching distal parts of the plant it exits the phloem and triggers silencing in the recipient cells. As is logical, this signal follows a sink-source pattern reminiscent of that followed by moving viruses (Voinnet, 2005). The mobile silencing signal has been shown to be siRNAs in N. benthamiana (Hamilton et al., 2002). Accordingly, sRNAs of all kinds have been found in phloem exudates of B. napus (Buhtz et al., 2008). Rigorous genetic experiments in A. thaliana have shown that the transgene-driven mobile signal targeting an endogenous gene is DCL4-dependent 21nt siRNAs. These siRNAs move cell-to-cell not in complex with AGO1, but require AGO1 in the recipient cells to efficiently trigger silencing. Moreover, leaf bombardment with artificial siRNAs suggested that these move as duplexes (Dunoyer et al., 2010). Also 24nt siRNAs are able to move cell-to-cell, but are not able to trigger PTGS (Dunoyer et al., 2007, 2005). However, another study proved that 24nt siRNAs can move systemically and trigger TGS in distant tissues (Molnár et al., 2010). Transgenic hairpin-driven DCL2-dependent 22nt siRNAs are able to move cell-to-cell and trigger PTGS (Dunoyer et al., 2007), although it remains unclear whether they can do the same systemically or not. Since 21nt and 24nt siRNAs are able to move systemically it is reasonable to assume that 22nt siRNAs should be capable of systemic movement, though it isn’t known to what extent they can mediate PTGS in systemic tissues. It has been shown that upon exiting the phloem, primary siRNAs can move up to 10-15 cells, after which they stop being able to mediate PTGS, either because movement entails a dilution effect or because of some intrinsic change in the siRNAs. However, in the presence of RDR6 and an siRNA-homologous template, the silencing signal can be amplified and move further (Himber et al., 2003). What factors, if any, are needed for cell-to-cell and systemic movement of siRNA duplexes remains an open question.
Given the extensive data gathered through transgenic reporter systems on the movement of siRNAs and its consequences, it is logical to assume that during viral infection the antiviral siRNAs produced by DCLs at the site of infection would, in addition to mediating intracellular AGO and RDR activities, move cell-to-cell and systemically ahead of the virus. This way, the yet-uninfected tissues could employ the inbound virus-derived siRNAs to mount a preemptive sequence-specific defense to stop the virus upon arrival. It has been shown that CymRSV, a (+)ssRNA virus, lacking ability to neutralize 21nt siRNAs can leave the infected tissues to move systemically but can’t exit the phloem once it reaches the
Figure III: Systemic antiviral RNAi in plants.
Schematic representation of non-cell autonomous, systemic antiviral RNAi. The primary RNAi response in infected tissues generates mobile siRNAs (1) which move cell-to-cell and systemically, priming a preemptive RNAi response in distant naïve tissues (2). Subsequently, when the virus exits the primarily infected tissues (3) and tries to colonise new ones, it faces a powerful and specific RNAi reaction (4).
22
systemic tissues (Havelda et al., 2003), presumably because of the ability of 21nt siRNAs to prime effective defenses before or upon arrival of the virus. However, this could also be explained as an unprimed reaction to the virus, which is instantly stopped upon arrival because it lacks a VSR. Unfortunately this is the only reported example of an antiviral strategy that is presumably widely employed by plants, and leaves us not only wondering if and how much it is used during a genuine virus infection, but also which are the mechanisms and factors involved in the perception of this virus-specific “immunogenic” signal and its implementation in the preemptive defensive reaction (Fig. III).
While RDR6 is strongly involved in the perception of the systemically mobile silencing signal in transgenic systems (Himber et al., 2003; Voinnet, 2005), presumably through its ability to amplify this signal in the recipient tissues, this cannot logically happen in viral infections. In fact, if the silencing signal moves ahead of the virus, the perceiving tissues have no template for RDR6-dependent generation of more siRNAs. However, upon arrival of the virus, RDR6 could use viral RNA to generate dsRNA for DCLs to make more antiviral siRNAs. These secondary siRNAs could mediate cell-autonomous silencing, “alert” the neighboring cells or, through further phloem loading, reach yet more systemic tissues. The factor/factors perceiving mobile antiviral siRNAs are not known, though it could be speculated that AGO1 is involved (Dunoyer et al., 2010).
The difficulty in studying the antiviral aspect of mobile RNAi is mostly determined by the fact that in most viral infections it is not possible to experimentally untangle the virus’ (i) ability to move, (ii) ability to block cell-autonomous RNA silencing and (iii) ability to block systemic mobile RNA silencing. In this work, thanks to a naturally occurring property of the viral protein P15, we managed to uncouple systemic from cell-autonomous RNAi in the context of a genuine viral infection. We hence provide evidence that mobile virus-derived 21nt siRNAs can strongly hinder viral systemic movement, at times completely abolishing it.
By now it is clear that a virus entering a cell is confronted with an adaptable, multi-layered, sequence-specific and aggressive defensive system, and that it cannot even hope to escape these defenses by moving to other cells and tissues, because it will be followed (or preceded!). In their struggle for survival viruses have come up with a myriad of tricks and
23
stratagems to neutralize RNAi, and scientists are barely starting to scratch the surface of this treasure trove.
III - Viral suppression of RNA silencing
Before going into the molecular details of viral suppression of silencing through especially adapted proteins, it is worth looking at the life cycle of a typical virus to identify some of the less specific and elaborate strategies through which it can evade RNAi.
The first and most obvious is the capsid. Whatever the number of strands and secondary structures of an RNA, a proteic shell will protect it from DCLs, AGOs and other nucleolytic enzymes. The second is the replication within membrane invaginations, forming partially enclosed viral factories that can act as protective alcoves, totally or partially inaccessible to the host’s defensive machinery. A third strategy could be, so to speak, to “fly under the radar”. Entailing low rates of replication to avoid full-blown RNAi reactions, this strategy could be employed by those viruses mentioned at the beginning of this chapter, that are present in low titers in their hosts but possibly over many generations. To state it differently, avoiding pathogenesis and an explosive rate of replication may be a valid means to evade RNA silencing. A fourth strategy could be to avoid triggering RNAi whenever possible by avoiding secondary RNA structures. From this point of view every secondary structure still present in a viral RNA genome after millennia of evolution (and many are described) could be presumed to be strictly necessary, or at least highly advantageous, to the viral life cycle. One case is reported where a non-coding secondary structure may act as a decoy for DCLs, diverting them away from the vital coding regions of the genome (Blevins et
al., 2011). Such “DCL-sponges” could be a widespread strategy. Additionally, although no
such example has been specifically described, it would be of great advantage to the virus to be able to quickly separate the (+) and (-) RNA strands after replication to avoid triggering RNAi. The helicase domain present in many viral replicases could be responsible for this. A fifth possible strategy, despite being speculative and not confirmed by experimental data, is worth considering here. A fair number of virus species are phloem-limited. These viruses can be found in many different families and genera. It is certainly possible that these viruses actively avoid exiting the phloem, or are for some reason unable to move and replicate
24
outside this tissue. Another explanation could be that they are confined to the phloem because they lack the ability to sufficiently suppress plant defenses to exit this tissue. This hypothesis entails that the phloem would be somehow lacking in antiviral defenses, possibly antiviral RNAi. A few pieces of experimental evidence, while not directly addressing this question, may point toward a phloem-specific RNAi complement. Results obtained by Parent et al., 2015 suggest that DCL2 is almost absent in companion cells. Some results presented in this work in Chapter 1 can be interpreted the same way. Parent and colleagues in the same study also show that transgenic phloem-generated 22nt siRNAs can induce RDR6/DCL4-dependent transitivity against an endogenous gene. Given the fact that the phloem is connected to the whole plant, mobile transitivity-inducing 22nt siRNAs could potentially have seriously detrimental plant-wide effects, so a strict control of DCL2 accumulation in this tissue wouldn’t be surprising. Furthermore, the aforementioned study by Havelda and colleagues clearly shows that a virus lacking the ability to neutralize 21nt siRNAs can survive in the cells surrounding the sieve elements but cannot move further. One interpretation of these pieces of evidence combined is that in the phloem very little or no secondary RNAs are made because of the very low levels of DCL2 and transitivity-inducing 22nt siRNAs, so during a viral infection only primary DCL4-dependent 21nt siRNAs are available to mediate RNAi, leading to a reduced response and consequent survival of viruses with weak RNAi-suppressing ability. An additional evolutionary advantage of this state of affairs could be postulated: by keeping an RNAi-poor environment and therefore easing the selective pressure on viruses whose RNAi-suppression capacities are scarce, plants would greatly reduce the likelihood of these viruses developing strong suppressors of silencing.
While being hypothetical, these possibilities are worth experimental investigation. Nonetheless, if the phloem is indeed a low-RNAi haven, phloem restriction could be a way for viruses to avoid a full RNAi response, although this would be more an unwilling consequence than an active strategy. Confinement to the phloem could have the further advantage of facilitating transmission to new hosts by phloem-feeding vectors.
In addition to these more general strategies, viruses have evolved proteins that actively target specific steps of RNAi. These viral suppressors of RNA silencing (VSRs) have been
25
extensively studied, and the most significant results of these investigations are described in the following review.
Feature Review
RNA
silencing
and
its
suppression:
novel
insights
from
in
planta
analyses
Marco
Incarbone
and
Patrice
Dunoyer
IBMP-CNRS,12rueduGeneralZimmer,67084StrasbourgCedex,France
Plants employ multiple layers of innate immunity to fight pathogens.For bothRNAandDNAviruses,RNA silencing plays a critical role in plant resistance. To escapethisantiviralsilencing-basedimmuneresponse, viruseshaveevolvedvariouscounterdefensestrategies, themostwidespreadbeingproductionofviral suppres-sorsofRNAsilencing(VSRs)thattargetvariousstages of the silencing mechanisms. Recent findings from in planta analyses have provided new insights into the modeofactionofVSRsandrevealedthatplantsreact totheperturbationofthesilencingpathwaysbroughtby viralinfectionbydeployingabatteryof counter-counter-defense measures.Aswellasdiscussingwhich experi-mental approaches have been most effective in delivering clear and unambiguous results, this review providesadetailedaccountofthe surprisingvarietyof offensive and defensive strategies set forth by both virusesandhostsintheirstruggleforsurvival.
RNAsilencingpathways
RNAsilencingisapaneukaryoticgeneregulation mecha-nism with fundamental implications in many biological processes.Itistriggeredbydouble-strandedRNA(dsRNA) andcausesasequence-specificshutdownoftheexpression ofgenescontainingsequencesidenticalorhighlysimilarto theinitiatingdsRNA.Inplants,RNAsilencingactsatboth theRNAandtheDNAlevels.Mechanismsofsilencingat the RNA level include mRNA cleavage or translational repression,whereas attheDNAlevelthey involve DNA and/orhistonemethylationandsubsequent transcription-al gene silencing (TGS)through heterochromatin forma-tionandmaintenance[1,2].
Allthesemanifestationsof RNA silencingrelyon the action of smallRNA (sRNA) molecules of 21–24 nt that originate from the processing of the dsRNA trigger by RNaseIII-likeenzymescalledDicer,orDicer-likeinplants. ThemodelplantArabidopsisthalianaencodesfour Dicer-likeproteins(DCL1,DCL2,DCL3,andDCL4),eachwith specializedfunctions[3].DCL1mainlycontributestothe production of miRNAs [1], whereas DCL4, DCL2, and DCL3 generate populations of 21-, 22-, and 24-nt short-interferingRNAs(siRNAs),respectively[4,5].
Upon processing, one strand of the sRNA duplexes generated by DCLs is incorporated into an Argonaute (AGO)-containingRNA-inducedsilencingcomplex(RISC) toguidesequence-specificinactivationoftargetedRNAor DNA. Most miRNAsandDCL4-dependent 21-ntsiRNAs loadintoAGO1toguidepost-transcriptional gene silenc-ing (PTGS) of target mRNAs [1]. These target mRNAs encode transcription factors required for plant develop-ment, as well asenzymes involved invarious metabolic andhormonalpathways[1].Upontheirincorporationinto AGO4, AGO6, or AGO9, DCL3-dependent 24-nt siRNAs act mostly in cis, to direct cytosine methylation and chromatin modifications at endogenous loci, including transposonsandrepetitivesequences,inaprocessknown asRNA-directedDNAmethylation(RdDM)[2].
AntiviralRNAsilencing
Besidesitsrolesindevelopmentalpatterningand mainte-nanceofgenomeintegrity,RNAsilencingalsoconstitutes theprimaryplantimmunesystemagainstviruses. Antivi-ralRNAsilencingistriggeredbydsRNAreplication inter-mediates or intramolecular fold-back structures within viralgenomes[6–8].TheseviraldsRNAsaremainly pro-cessedbyDCL4oritssurrogateDCL2,toproduce21-or 22-ntvirus-derivedsmallRNAs(vsRNAs),respectively[9,10]. OptimalproductionofvsRNAalsorequiresdsRNA- bind-ing proteins(DRBs)such asDRB4,whichfacilitates the synthesisofDCL4-dependentvsRNAfromRNAandDNA viruses[11,12].vsRNAsaresubsequentlyrecruited, main-lybyAGO1andAGO2,todirectPTGSofviralRNAaspart of antiviral RISCs (Figure 1) [12–19]. In DNA virus-infected plants (gemini- and pararetroviruses), a large amountofDCL3-dependent24-ntvsRNAsisalsoproduced
[9,20]. These 24-nt vsRNAs direct cytosine methylation and chromatin condensation of nuclear viral episomes andminichromosomestodampenviraltranscription,most likely throughAGO4activity(Figure 1)[21–23].The in-versecorrelationbetweenthelevelofviralDNA methyla-tionandtheseverityofviralsymptomssuggeststhatplant defense against DNA viruses also relies on the RdDM pathway.
To cope with the high pace of virus replication and movement, plants have also evolved means to amplify theantiviralsilencingresponse.Thisoccursthrough pro-ductionofso-calledsecondaryvsRNAs,asopposedtothe primary vsRNAs which are generated directly from the structural features or replication intermediates of viral
Review
1360-1385/$–seefrontmatter
!2013ElsevierLtd.Allrightsreserved.http://dx.doi.org/10.1016/j.tplants.2013.04.001
Correspondingauthor:Dunoyer,P. (patrice.dunoyer@ibmp-cnrs.unistra.fr).
RNA.ThesesecondaryvsRNAsareproducedthroughthe activity of cellular RNA-dependent RNA polymerases (RDRs),whichconvertsingle-strandedRNA(ssRNA)into newdsRNA substratefor processing byDCLs(Figure1)
[24].InArabidopsis,RDR1,RDR6,and,toalesserextent, RDR2have beenimplicatedinthis amplification step of vsRNAaccumulation, which sometimes accountsfor the largestbulkoftheantiviralsRNAproduced[12,25–29].
Moreover,antiviralRNAsilencingcanalsospreadfrom thesiteofinitiationtothesurroundingtissues[30,31].In plants,thenatureofthemobilesilencingnucleicacidsthat convey sequence specificity has beenunambiguously as-cribedtosiRNA[32–34].Thisnon-cellautonomousaspect of RNA silencing represents the systemic arm of this antiviralreaction,wherebytransmissionofmobilevsRNA ahead oftheinfectionfront primesantiviral silencingin cellsthatareyettobeinfected(Figure1).Consequently, replicationormovementofthepathogenintothosecellsis delayedorprecluded[35–37].
ViralsuppressorsofRNAsilencing
Oneofthemostcompellingpiecesofevidencesupporting RNAsilencingasamajorantiviraldefensemechanismin plantsistheobservationthatmost,ifnotall,phytoviruses have evolvedmeans to counteract,attenuate, or escape this response [38].Among these, theproduction of viral suppressors of RNA silencing (VSRs)is by far themost widespreadviralcounterdefensestrategyemployed.VSRs are highly diverse in sequence, structure, and activity within and across virus families, suggesting that their acquisition occurred through rapid evolutionary conver-gence as a mandatory adaptationto the RNA silencing-based immune response. In agreement with this, VSR expressionisoftenaprerequisiteforvirusmultiplication and systemic host infection in both plants and insects
[10,27,28,39].AlthoughVSRshavebeenshownto target manystagesoftheantiviralsilencingpathway(Figure1), theirmodesofactionareusuallyclassifiedintothreebroad categories:(i)bindingtolongdsRNAresultingininhibition
AAA Replica!on Geminivirus Pararetrovirus DCL2 DCL4 DCL3 Viral mRNAs dsRNA intermediates DCL2 24nt AGO4 DNA/histone methyla !on AAA 60S 40s AAA An!viral RISC Transla!onal inhibi!on Transcrip!on DCL4 AGO1 AGO1 AGO1 21nt 21nt DCL4 21nt RDRs P AGO1 Immuniza!on AGO1 Immuniza!on Amplifica!on Cleavage Dicing Cell-to-cell spread Cell-to-cell spread Dicing Viral
mRNAs mRNAsViral Dicing
P6 βC1 2b V2 P19 P21 P1 2b P19 P21 P P38 AGO1 P0 Autophagosomes
TRENDS in Plant Science
Figure1.AntiviralRNAsilencinganditssuppressionbyvirus-encodedsilencingsuppressorsinplants.AntiviralRNAsilencingistriggeredbydouble-strandedRNA (dsRNA)replicationintermediatesorintramolecularfold-backstructureswithinviralgenomesthatareprocessedintovirus-derivedsmallRNAs(vsRNAs)byRNaseIII-like enzymescalledDicer-likeproteins(DCL4,DCL3,andDCL2).ThesevsRNAsareloadedintoArgonaute(AGO)-containingRNA-inducedsilencingcomplexes(RISCs)toguide translationalinhibitionand/orslicingofviralRNA.CleavedviralRNAsarealsousedbycellularRNA-dependentRNApolymerases(RDRs)andtheircofactorstoamplifythe RNAsilencingresponsethroughproductionofmoredsRNAsubstratesforDCLprocessing.BothprimaryandsecondaryvsRNAsalsohavethepotentialtomoveto neighboringcellsthroughplasmodesmatatoprompttheantiviralsilencingresponse.ForDNAviruses,alargeproportionofDCL3-dependent24-ntvsRNAsarealso produced.These24-ntvsRNAsdirectDNAand/orhistonemethylationoftheviralDNAgenomes.ViralsuppressorsofRNAsilencingcaninhibitvariousstagesofthis pathway,therebypreventingdicing,vsRNAloading,RISCformationoractivity,amplification,andmovement.ThestepstargetedbysomeoftheVSRsdiscussedinthis review(P19,P21,P38,P1,2b,P0,P6,V2,andbC1)aredepicted.Abbreviation:P,plasmodesmata.
Review TrendsinPlantScience July2013,Vol.18,No.7
